Behavior Adjustment Training
(See below for more resources)
Warning: this is a dog geek page! Done right, BAT is like a miracle. I actually have gotten goosebumps at the end of a session. Behavior Adjustment Training, or BAT, uses environmental consequences to reward socially acceptable choices. In a nutshell, BAT gives the dogs a chance to learn to control their environment through peaceful means. It’s very empowering to the dog, in a good way.
BAT uses functional and bonus rewards, combined with clicker training and systematic desensitization to help dogs make better choices in an error-free way.
The things that you find obnoxious or scary (barking & lunging, for example) serve a purpose for the dog. That is the idea behind what behavior researchers call Functional Analysis.Â Imagine you hated your high-stress job but loved the pay. It leaves you exhausted and panting, but you enjoy the reward at the end of it. Given the chance to earn your same great paycheck by doing something easier that you liked more, wouldn’t you do it?
BAT is similar to a method used with humans, called Functional Communication Training. Students learn to communicate their needs instead of using aggression or other problem behavior. Sound familiar?
Functional Communication Training, used with children with developmental disabilities, is one of the inspirations for BAT, along with the best parts of clicker training, systematic desensitization, Constructional Aggression Treatment, the Two-Reward method, Treat & Retreat, click and retreat by Alexandra Kurland, and more. Functional rewards can fall into the quadrant of positive reinforcement (like ‘real life rewards’) and negative reinforcement (like relief from social contact). The upcoming book on BAT goes into that in more detail.
(See transcript of video here- it helps!)
Functional rewards are not just for aggression or fear, but eliminating distress that leads to reactivity (including frustration, anger, and fear) is the main thrust of BAT. BAT uses Differential Reinforcement of Alternate behaviors (DRA) with environmental rewards: the ones the dog is already working to get. BAT also uses ‘Bonus Rewards’ when necessary for motivation – food, toys, etc. Â The key to BAT working quickly is low stress, never leaving the dog in the ‘deep end of the pool’ or punishing them for incorrect choices.
First, here’s why I’m calling it BAT.Â The goal is a well-adjusted dog and behavior is all we’re really trying to adjust. When you’re done, you have the same trigger as before, and the same consequence, but a new behavior. The core idea that I love from Functional Analysis is the idea of using the reward that the dog is working to earn to reinforce a new, more ‘appropriate’ replacement behavior. So if the dog is already getting paid for barking at the mailman by having him leave, the new set-up is to pay the dog for calm behavior by having a pseudo-mailman leave as a consequence for good behavior.
With the Antecedent -> Behavior -> Consequence model, you have something like: Antecedent -> Behavior 1 -> Consequence being replaced by Antecedent -> Behavior 2 -> Consequence (Think of ‘antecedent’ as a cue or something that signals the behavior is about to be paid for.) So Mailman arrives -> Barking -> Mailman leaves is replaced by Mailman arrives -> dog turns away -> Mailman leaves I love that part. It means that the environment itself, which created the problem, will now start rewarding your dog.
You will need to have several friends practice being the mailman before it works for real. The bigger picture of BAT is that it includes real-world positive reinforcement, too: Mom comes home -> jumping -> petting/attention is replaced by Mom comes home -> sitting -> petting/attention Not that that’s new, but I love that the BAT model fits in with what we already do! The behavior is what we adjust, within the environment that provides the antecedent and consequence. That’s why I’ve called it Behavior Adjustment Training. Now to get that behavior change, we have to make sure that either Behavior 1 -> Consequence is a chain that gets broken, as we do when we ignore jumping, or we work on more errorless learning. What I mean by that is that we start with an antecedent that is different enough from the usual antecedent that the dog doesn’t do the naughty Behavior 1.
It gradually morphs over time until the picture becomes Antecedent -> Behavior 2 -> Consequence, as we wanted. Mailman really, really far away -> dog turns away -> Mailman leaves … Mailman really far away -> dog turns away -> Mailman leaves …and so on, until we have Mailman arrives -> dog turns away -> Mailman leaves
In that example, the dog is still getting rid of the mailman. With reactivity to strangers or other dogs, the idea is that the dog learns that it’s safe to approach. Curiosity can blossom, which then leads to trust and sociability. For example, my dog who was afraid of people now leans in for petting from strangers, because he learned that he could just walk away when he felt a little nervous.
Here is the BAT protocol for aggression or fear in its simplest form.
- Expose: Start sub-threshold and remain below the dog’s threshold as you increase stimulus intensity a little (e.g. student dog moves closer to stooge (a.k.a. helper/decoy/actor), stooge moves closer, or both).Â If dog is getting worse instead of better, abort.
- Wait for or manufacture acceptable alternative behavior in a non-aversive way. Be sure to take very small behaviors, like blinks or head turns. Waiting is preferred, whenever possible.
- Mark using a verbal Yes or a clicker.
- Functional Reward: Decrease stimulus intensity (but not to zero). ex. Student gets to walk 20 feet away, but stooge remains in view.
- (optional) Give a treat or toy as a Bonus Reward (best to use on Walks, not usually necessary or desired for set-ups).
This need not be a big, long set-up. It can just be a single repetition, in the middle of doing something else, like TTouch ground work or passing by a gardener in front of her home. Whenever the dog is in a slightly stressful situation and they do a nice appropriate signal (instead of aggression) you could mark with “Yes” and retreat with them as the reward. What’s very cool about BAT for reactivity is that the dog starts to actually become friendly to the decoy, even though all they wanted, at the beginning, was more distance. It’s as if they now have the locus of control and the world makes sense again. Their peaceful choices are controlling their environment, and having an internal Locus of Control feels good!
Unlike counterconditioning, which assumes that emotions drive behavior, BAT acknowledges the theory that behavior and its consequences can also drive emotions. I think both are probably true, so I use both CC/DS and BAT, in combination or in different circumstances. Key points to remember when doing BAT for aggression or fear:
- Above all, stay sub-threshold. Take frequent breaks, especially when the dog asks for them.
- If the dog becomes agitated (breathing heavier, looks like he might have an outburst), interrupt the dog or counter-command (give the cue for an easy incompatible behavior) and then decrease the stimulus intensity (for example, walk away from the decoy or have them go away, whichever is least reinforcing). If the dog has an outburst and you just stand there, you’re wasting time, because he’s more sensitive to the stimulus for your next trials, and the rate of reinforcement goes way down. Barking/lunging is also hard on everybody else’s stress level, including the stooge and the student dog’s owner.
- Shape for a nice big set of clear cut-off signals. Don’t go for variety too soon. The student dog (or horse or goat or person or whatever) should experience rapid success. You can repeat this later to look for variety. Replacement behaviors include: blinking, jaw loosening (being able to pant again), looking away, turning away, ground sniffing, air sniffing, tail carriage getting looser.
- Reinforce calmly gathering information with allowing the dog to continue doing so, or with quiet praise.
- I really like having the student dog able to move, rather than being tethered, and able to walk away from the situation (safely leashed or behind a barrier, but not tethered to a wall). If the student requests a break by checking out, she gets one. Dogs should have control of their exposure to the trigger during the session (your job is often just to keep them from getting so close that they go over threshold).
- For most dogs, especially fearful dogs, I prefer for them to retreat from the Scary Monster as their reward when they’ve done an acceptable alternative behavior, versus only having the monster leave. Fear is the emotion of ‘get me out of here!’ so it makes more sense to me that the student dog gets to leave versus chasing off the bad guy. You do need to have trials where the stooge approaches and retreats, but I think the bigger reward for a fearful dog is being able to leave. So even if the stooge approaches the student dog, the student dog still walks away as the reward, simultaneously or just after the other dog leaves. Even dogs who are offensively aggressive (angry, territorial, whatever you want to call it)Â benefit from learning how to just walk away. Lateral/tangential retreats work well for them, at first.
- If the student dog barks on retreat, continue to retreat, but remember that next time you’ll need to tone down the stimulus (a bit farther away, less motion or whatever).
- It is totally reasonable to pre-train some of these behaviors via clicker training without the decoy there.
- What about regular walks between BAT sessions? Any time you encounter triggers that will hold still (dogs behind fences, people out gardening, etc.), you can do BAT by approaching and retreating based on the dog’s behavior.
- On walks where you don’t control the triggers, follow the steps on this handout. The core of BAT is simply that you use environmental rewards when you can, to change A -> B1 -> C to A -> B2 ->C. When you can’t use environmental rewards, at least give the dog a chance to earn reward somehow, even by just looking at the trigger!
- You don’t need to do a marathon session to complete switchover (where you can tell trust has built and the dog is seeking social contact), although you can.
- The first session of BAT may take a long time to switchover (the point where the dog wants to be near the stooge versus only leaving), the second will take less time, and so on. Eventually it’s like the real world, with no time to switchover. That’s how you know the dog has been rehabilitated (assuming it’s a generalized response). If you don’t get to switchover in the first session, that’s fine. Just try to end each session with some down-time, hanging out at whatever distance feels comfortable to the dog. You can use the same stooge again and work up to socializing with them, unless that’s not safe.
- Remember that we’re teaching new skills to negotiate with the Scary Monster, so the first several times that you do this will take more time than, say, counterconditioning, or simply parallel walking. But as I said before, the time-to-switchover shrinks rapidly. And now the dog has a new set of skills.
- This involves some negative reinforcement, because ‘relief’ is the treat. While I would never apply pain in order to remove it, this kind of negative reinforcement is very natural, and done right, need not be too aversive, and we need not apply anything aversive to the dog in order to take it away as a reward. In many cases, their own species is the unwanted stimulus and in most cases, they are walking themselves up to the approach line. They will be exposed to other dogs all the time anyway. They are already learning behaviors anyway. We are just teaching them different behaviors, using their natural environment.
- The marker signal of “yes” becomes a classically conditioned safety cue for the dog, because it has signaled the end of a trial and a retreat.
- Official BAT site, FunctionalRewards.com
- Available in the online store: 1.5-hour Intro to BAT DVD, “Organic Socialization: BAT for Aggression & Fear in Dogs (a DVD made for home viewers, with 4 hours of footage), and a Full-day BAT seminar DVD. You can also pre-order the book Behavior Adjustment Training: BAT for Aggression, Frustration, & Fear in Dogs.
- Join FunctionalRewards, a Yahoo! discussion group and creative space for people interested in BAT and related or compatible methods.
- Read more posts about BAT.
- Read this Q & A about BAT on the Rewarding Dog Behavior Training forum (Feb, 2010).
- Read some comments or leave your own on the BAT comment page.